beta(1)- and beta(2)-adrenoceptor responses in cardiomyocytes derived from human embryonic stem cells: comparison with failing and non-failing adult human heart.

BACKGROUND AND PURPOSE: Characterization of human embryonic stem cell-derived cardiomyocytes (hESC-CM) in relation to adult myocytes is essential for their future use in transplantation or as a model system. The beta-adrenoceptor pathways, which are known to be effective early in hESC-CM development, are of major importance because of their control of rate and force of beating, arrhythmia generation and apoptosis/necrosis. We have therefore performed detailed pharmacological analysis of the beta-adrenoceptor responses in developing hESC-CM. EXPERIMENTAL APPROACH: hESC-CMs were differentiated from H7 ESCs and studied up to 79 days of differentiation. Rate of beating and time course of contraction and relaxation were measured in superfused preparations. KEY RESULTS: Responses to the mixed beta(1)- and beta(2)-adrenoceptor agonist isoprenaline were evident from day 10 to day 79. Stability of the responses during an application, for repeated applications on the same experimental day and over the time of development, was determined. Concentrations for half-maximal response (12.9 nM) were similar to those from adult human heart, but closer to those obtained from failing rather than normal ventricle. Acceleration of both contraction and relaxation was quantitatively similar to that in adult ventricular myocytes, as was sensitivity to muscarinic inhibition. Use of specific antagonists showed that both beta(1)- and beta(2)-adrenoceptors contributed to contractile responses, as seen with adult myocytes. CONCLUSIONS AND IMPLICATIONS: These data show the compatibility of hESC-CM with adult human myocardium in terms of beta-adrenoceptor response. The experiments described here also confirm the utility of the hESC-CM preparation for detailed pharmacological analysis.

Constitutive expression of non-bone/liver/kidney alkaline phosphatase in human osteosarcoma cell lines.

Alkaline phosphatase (ALP) plays an important role in bone mineralization; high levels in differentiated osteoblasts allows their identification easily in vitro. It is generally assumed that the activity of ALP in osteosarcoma-derived cell lines commonly used in studies of bone cell biology is exclusively due to the bone/liver/kidney (BLK) isoenzyme. However, we noted that two human osteosarcoma cell lines, U-2 OS and U-393 OS, predominantly expressed a truncated 1.8 kb mRNA for BLK-ALP. This observation stimulated further investigation upon the ability of ALP to form functional protein. We found that, unlike the BLK-ALP of the Saos-2 osteosarcoma cell line, the activity of U-2 OS ALP was thermostable, unaffected by L-homoarginine and levamisole, but inhibited by L-phenylalanine; these properties are characteristic of the placental and/or placental-like (PL-/PL-like ALP) isoenzymes which are 98% homologous at the amino acid level. In the U-393 OS cell line, which expresses the normal-sized 2.5 kb mRNA in substantially higher levels than that produced by U-2 OS cells, the ALP activity had kinetic properties very similar to that produced by the Saos-2 line for all criteria tested. The HOS osteosarcoma cell line (also known as TE-85), which express the normal-sized 2.5 kb BLK-ALP mRNA only, exhibited ALP activity with kinetic properties of both the BLK and PL-/PL-like isoenzymes. The three test lines, U-2 OS, U-393 OS and HOS, produced PL-/PL-like ALP mRNA and protein constitutively, and levels of these increased in cells treated with 1 microM dexamethasone. However, dexamethasone treatment of cells did not alter the types of ALP isoenzyme expressed. Thus our results show that, like Saos-2 cells, U-393 OS cells produce active BLK-ALP exclusively, whereas U-2 OS cells produce PL-/PL-like ALP only, and the HOS cell line produces both. Our findings have important implications for phenotypic characterization of various human osteosarcoma cell lines, and suggest that the production of PL-/PL-like ALP may be a more common occurrence in osteosarcomas than was originally thought.

Parathyroid hormone, but not prostaglandin E2, changes the shape of osteoblasts maintained on bone in vitro.

Parietal bones from 2-week-old rats were dissected free from the sutural regions, dura mater, and periosteum, leaving the surface covered with osteoblasts and some osteoclasts. Prostaglandin (PG) production by these "stripped" bones under basal conditions and after exposure to parathyroid hormone (PTH) was measured by radioimmunoassay of the culture medium (minimum essential medium with or without added 10% heat-inactivated fetal calf serum). Cultured specimens were examined by scanning electron microscopy for changes in osteoblast length, orientation, ruffling, and overlap. As demonstrated previously, PTH caused the osteoblasts to elongate, align, and show fewer ruffles compared to controls. PTH increased PG synthesis by the stripped bones. Indomethacin inhibited PG formation but did not affect the osteoblast shape change. PGE2, indomethacin, or both drugs together had no discernible effect on any morphologic features. These findings indicate that PGE2 does not change osteoblast shape and that the cell shape change with PTH is not mediated by endogenous prostanoids.

Activation of NF-kappaB in human osteoblasts by stimulators of bone resorption.

Several bone resorptive stimuli affect osteoclasts indirectly by modulating the production and release of osteoblastic factors. Using electrophoretic mobility shift assays, we found that not only tumour necrosis factor-alpha (TNF-alpha) but also interleukin-1beta and parathyroid hormone (PTH) caused dose and time-related increases in nuclear factor kappaB (NF-kappaB)-DNA binding in Saos-2 human osteoblastic (hOB) cells. Activation of NF-kappaB by TNF-alpha was reproduced in primary hOBs. In contrast, consistent with their previously reported lack of response to steroid hormones, Saos-2 cells did not respond to 1,25-dihydroxyvitamin D(3). We suggest that NF-kappaB activation in osteoblastic cells constitutes an important pathway in osteoblast-mediated resorptive signalling.

Spectrum of osteoblastic differentiation in new cell lines derived from spontaneous murine osteosarcomas.

Cell lines were established from three spontaneous osteosarcoma and one fibrosarcoma of aging mice. They were studied for tumorigenicity, osteoblastic features, and other in vitro cellular characteristics, by a combination of histological, morphological, biochemical, and molecular approaches. It was found that all cell lines formed tumors in vivo, whereas in vitro, only the fibrosarcoma-derived cell line grew efficiently in soft agar. Three out of the four cell lines produced mouse endogenous retroviruses, but none were classical sarcoma viruses. Type I collagen was expressed by all the cell lines, as was another extracellular matrix protein, osteonectin. The osteosarcoma-derived cell lines, however, exhibited different degrees of osteogenic differentiation. Only one line (OSA), and its clonal subline (1G11), consistently gave rise to mineralized tumors after transplantation into syngeneic mice, and these cells expressed high levels of alkaline phosphatase and bone-specific osteocalcin mRNA in vitro. Expression of these biochemical markers of osteoblasts occurred to a lesser extent in a second line (OSC) and was undetectable in the third line (OSB). The clonal 1G11 cell line exhibits the phenotype of a fully mature osteoblast and thus may serve as a particularly useful model for studies of bone cell function and regulation. Studies of cells which display a wide spectrum of osteogenic potential may further our understanding of the mechanisms involved in bone cell differentiation and tumorigenicity.

The effects of inhibitors of cysteine-proteinases and collagenase on the resorptive activity of isolated osteoclasts.

The effects of specific inhibitors of cysteine-proteinases ((Z-Phe-Ala-CHN2: benzyloxycarbonyl-phenyl-alanyl alanyl diazomethane and E-64: trans-epoxy-succinyl-L-leucylamido(4-guanidino)-butane) and collagenase and collagenase ((Cl-1: N-(3-N-benzyloxycarbonyl amino-1-R-carboxypropyl)-L-leucyl-O-methyl-L-tyrosine N-methylamide) have been tested on the osteoclastic resorption of dentine. Chick osteoclasts were cultured in the presence or absence of 12.5 microM Z-Phe-Ala-CHN2, 40 or 60 microM E-64, or 40 or 100 microM Cl-1 for 1 or 2 days. In addition, osteoclasts were cultured on oyster shell calcitostracum with or without 12.5 microM Z-Phe-Ala-CHN2. Specimens were studied by light microscopy to count cells and resorption features and by scanning electron microscopy (SEM) stereophotogrammetry for the measurement of the depths, plan-areas and volumes of resorption pits. The numbers, depths and volumes (but not the plan-areas) of the resorption pits in dentine were significantly reduced by Z-Phe-Ala-CHN2 and E-64. Thus, for a given plan-area, the volumes and the depths of resorption pits were smaller in these experimental groups compared with control dentine specimens. The overall inhibition of resorption was at least 75%. Cl-1 did not have this inhibitory effect on the numbers or sizes of resorption pits in dentine. When the oyster calcitostracum was used as a substrate for the osteoclasts, Z-Phe-Ala-CHN2 did not reduce the numbers or volumes of pits, but increased the plan-areas and prevented the formation of deeper pits.(ABSTRACT TRUNCATED AT 250 WORDS)

Initial observations on the effect of medium composition on the differentiation of murine embryonic stem cells to alveolar type II cells.

The pluripotency and high proliferative index of embryonic stem (ES) cells make them a good potential source of cells for tissue engineering purposes. We have shown that ES cells can be induced to differentiate in vitro into pulmonary epithelial cells (type II pneumocytes) using a serum-free medium designed for the maintenance of mature distal lung epithelial cells in culture (SAGM). However, the resulting cell cultures were heterogeneous. Our aim in this study was to attempt to increase pneumocyte yield and differentiation state by determining which medium components enhance the differentiation of pneumocytes and modifying the medium accordingly. Quantitative RT-PCR was used to measure changes in the expression of a type II pneumocyte-specific gene, surfactant protein C (SPC), in response to alterations in the cell culture medium. Results suggested that most individual SAGM growth factors were inhibitory for type II pneumocyte differentiation, with the largest increases in SPC expression (approximately threefold) being observed upon removal of retinoic acid and triiodothryonine. However, large standard deviations occurred between replicates, illustrating the highly variable nature of ES cell differentiation. Nevertheless, these observations represent an initial step towards achieving directed differentiation of pneumocytes from stem cells that could lead to their purification for tissue engineering purposes.

Survival and resorptive activity of chick osteoclasts in culture.

Previous studies have shown that osteoclasts obtained from chopped bones resorb surrogate calcified tissue substrata in vitro. These cultures contained all bone and marrow cell types pooled together. We have now parted the marrow from the bone and cultured the cells from the two fractions separately: on both resorbable substrates and on plastic in order to test their longevity in culture and ability to resorb following trypsinisation. Marrow-fraction, bone-fraction or whole bone derived cells were harvested from prehatch chick long bone shafts after removing the periosteum; seeded on sperm whale dentine (SWD) slices or plastic dishes and cultured continuously, or trypsinised and reseeded on to fresh substrata at weekly or half-weekly intervals. Observations were made by light microscopy and SEM. Many multinucleate cells were observed in the marrow fraction immediately after settling, deriving presumably from poorly adherent osteoclasts, next to bone, which had not been resorbing at the time of harvesting. By three days in culture on plastic, multinucleate cells were very large both in terms of plan extent and nuclear number: cell fusion occurred between osteoclasts and between osteoclasts and small, round uninuclear cells. SWD was extensively resorbed. The adherence of the osteoclasts was greater (a) to plastic upon trypsinisation than that of the other cells; and (b) to SWD than to plastic, particularly if the cells were resorbing. Trypsinised cells regained their resorptive capacity after seeding on to new SWD, but only for 1 or 2 treatments. Bone derived cells were similar to the marrow cultures, except for the much higher proportion of other bone cell types. Trypsinisation caused a higher proportional loss of multinucleate cells from both SWD and plastic. Resorption was still occurring at 6 weeks in all cultures. A wide diversity existed in the shapes, depths, plan areas and volumes of the resorption pits.

The resorption of biological and non-biological substrates by cultured avian and mammalian osteoclasts.

Mammalian and avian osteoclasts were isolated mechanically from long bones, seeded on to either untreated, unmineralized, anorganic or surface-demineralized mammalian dental tissues, and cultured for 1-6 h or up to 9 days in medium with added serum (10% heat-inactivated FCS). All substrates showed Howship's resorption lacunae which varied in detail with the composition and structural organization of the tissue. There was no species or substrate specificity. Osteoclasts also adhered, spread, migrated and resorbed in the absence of serum. In addition, osteoclasts resorbed avian egg shell and mollusc shell containing calcite and aragonite. When given the opportunity, osteoclasts are thus biochemically competent to resorb a much wider range of substrates than they usually do in vivo. Access to the substrate and attraction or deliverance of osteoclast precursors to it must be curtailing factors in in vivo resorption.

Motility and resorption: osteoclastic activity in vitro.

Rabbit osteoclasts (OCs), separated mechanically from long bones, were seeded on to glass or plastic substrates or slabs of sperm whale dentine (SWD). Cells were cultured in MEM with 10% FCS with or without added salmon calcitonin (SCT) at dosages of 0.001, 0.1 and 1 IU/ml. Although most rabbit SCT-treated OCs on the non-biological substrates showed inhibition of peripheral ruffling activity and motility at dosages that stop rat OC movement, resorption still occurred on the dentine. Thus such inhibition is unreliable as a general indicator for resorptive capability. Resorption lacunae were observed at all times from 6 h onwards. Using stereophotogrammetric techniques, the following minimum values were obtained from 24 h cultures: highest hourly rate of resorption of dentine for single OC, 570 micron3/h; average rate 165 micron3/h; mean total volume dentine removed per Howship's lacuna complex, 3,885 micron3; average value for plan area of surface attacked per OC, 1,450 micron2.

Monocyte-enriched cells on calcified tissues.

Monocyte-enriched human blood cells seeded on to sperm whale dentine and cultured for up to 20 days failed to produce any morphological signs of resorbtive activity, although multinucleate giant cells were formed. In contrast, preparations containing known osteoclasts derived from bone resorbed the same substrate within hours.

Myocardial tissue engineering: a review.

Myocardial tissue engineering, a concept that intends to overcome the obstacles to prolonging patients' life after myocardial infarction, is continuously improving. It comprises a biomaterial based 'vehicle', either a porous scaffold or dense patch, made of either natural or synthetic polymeric materials, to aid transportation of cells into the diseased region in the heart. Many different cell types have been suggested for cell therapy and myocardial tissue engineering. These include both autologous and embryonic stem cells, both having their advantages and disadvantages. Biomaterials suggested for this specific tissue-engineering application need to be biocompatible with the cardiac cells and have particular mechanical properties matching those of native myocardium, so that the delivered donor cells integrate and remain intact in vivo. Although much research is being carried out, many questions still remain unanswered requiring further research efforts. In this review, we discuss the various approaches reported in the field of myocardial tissue engineering, focusing on the achievements of combining biomaterials and cells by various techniques to repair the infarcted region, also providing an insight on clinical trials and possible cell sources in cell therapy. Alternative suggestions to myocardial tissue engineering, in situ engineering and left ventricular devices are also discussed.

Inhibition of osteoclastic motility by prostaglandins I2, E1, E2 and 6-oxo-E1.

We have recently found that calcitonin (CT), a hormone which inhibits osteoclastic bone resorption, completely abolishes the normally intense cytoplasmic motility of isolated osteoclasts. Here we report that prostaglandin (PG)I2, PGE1, PGE2 and 6-oxo-PGE1 cause an identical change in behaviour to that induced by CT. The order of potency was PGI2 greater than PGE1 greater than 6-oxo-PGE1 greater than PGE2. We found that, unlike CT which causes prolonged immotility in osteoclasts, the effect of these PGs was transient. The transient nature of the inhibition was not caused by their decay or inactivation, nor was it due to production in the cultures of a stimulator of osteoclast motility. Osteoclasts refractory to one PG were also less sensitive to the others, but showed no loss of sensitivity to CT, suggesting that the PGs share a common receptor system, distinct from that for CT. The PGs, like CT, appear to operate by increasing the cyclic AMP level in osteoclasts. The identical nature of the response of osteoclasts to PGs and CT, and the shared use of cyclic AMP as second messenger, suggest that the PGs, like CT, act directly on osteoclasts to inhibit bone resorption by these cells. Osteoblasts are known to make PGs, and we suggest that osteoblasts make them as agents of the local control of osteoclastic bone resorption. Paradoxically, when PGs are added to bone in organ culture they stimulate bone resorption. Like PTH they increase osteoblastic cyclic AMP levels, and the effect of adding PGs to bone may be a transient direct inhibition of osteoclasts followed by a sustained PTH-like stimulation of osteoclasts through osteoblasts. This mechanism may account for the bone resorption seen in inflammatory and malignant disease.

The effect of prostaglandin I2 and 6a-Carba-PGI2 on the motility of isolated osteoclasts.

We have recently found that calcitonin (CT), a hormone which inhibits osteoclastic bone resorption, completely abolishes the normally intense cytoplasmic movement of isolated osteoclasts. We have also found that prostaglandin (PG)I2 causes an identical change in behaviour. In this paper we extend our investigations into the mode of action of PGI2 and a stable analogue, 6a-Carba-PGI2. We found that, unlike CT which causes prolonged immotility in osteoclasts, the effect of PGI2 and 6a-Carba-PGI2 were transient. Our results suggest that the transient nature of the inhibition was neither caused by inactivation of these compounds, nor was it due to production in the cultures of an osteoclastic stimulator. CT, PGI2 and 6a-Carba-PGI2 all appear to operate by increasing the intracellular cyclic AMP level. We found no refractoriness to either CT, dibutyryl cyclic AMP or 8-bromo cyclic AMP, and neither PGI2 nor 6a-Carba-PGI2 affected the sensitivity of osteoclasts to CT or dibutyryl cyclic AMP. This implies that refractoriness of osteoclasts to PGI2 and 6a-Carba-PGI2 develops at some stage in the interaction between PG and cell proximal to cyclic AMP production. We also found that there was cross-tachyphylaxis between PGI2 and 6a-Carba-PGI2, and this suggests that these two compounds share a receptor site on osteoclasts.

The interface of cells and their matrices in mineralized tissues: a review.

The interface between cells and matrices in mineralized tissues formed in vivo has been studied mainly by looking at the matrix surface, which is easily prepared, and not at the cell surface, which presents problems. Vertebrate calcified tissues range from being acellular to highly cellular, but for all the tissues the formative cells lay down and organise a cell-specific matrix, although this may be deposited initially on a different tissue-type. The formation of hard tissues is a group activity of many cells; resorption is the province of one cell, though it may be controlled by others in the vicinity. Cell-matrix interfaces that develop in vitro have also mainly been studied at the matrix side. The main difficulty with in vitro studies of hard tissue interfaces is that the cells do not have the same activity or even cellular functions as they had in vivo under the complex control of physiological regulation. The question of osteoblastic osteoclasis falls into this category. It is possible to provide new substrata for both formative and resorptive hard tissue cells to test for the interaction between the cells and the 'matrix' on to which they are seeded. The changing cell-matrix interface may also be modelled using computer simulation of osteoclastic movement across a substrate based on known patterns exhibited by other cell types in vitro. Comparison with the shapes of complex resorption pits shows a surprising match. This suggests that the track of the osteoclast due to cell motility and the bone resorptive mechanism resulting in pits along that track are likely to be separately controlled phenomena.

Optical and scanning electron microscopy in the single osteoclast resorption assay.

The present studies relate to the single or isolated osteoclastic resorption function assay which we introduced in 1983 to overcome objections to assays based upon measurements of calcium release from bones, in which it was never strictly controlled whether the mechanism involved the destruction of bone with the formation of classical Howship's lacunae. The method may prove to be quite popular in the near future and has already been adopted by other research groups. In previous work, we had utilised stereophotogrammetry of scanning electron micrographs to measure the depth, volume and other parameters of the individual lacunae. However, increasing experience with the method has suggested that we can await a wide range of biological variability in single cell function in any one experiment. We have therefore tested other methods from which data could be obtained more rapidly to permit a better statistical analysis, albeit with reduced accuracy, of each resorption complex. The main aim of the studies reported here was to evaluate various methods of optical and scanning electron microscopy that can be used for the visualization of osteoclasts and their associated resorption lacunae generated in vitro in slabs of dentine and bone. Optical microscopy was found to be complementary to SEM, enabling vital microscopy of unstained and stained cells. In particular, oblique illumination LM and tandem scanning reflected LM (TSRLM) proved to be of paramount value for this purpose. Fixed coated specimens could be most rapidly scanned for resorption lacunae using darkfield reflected LM or TSRLM.

Variation in the sizes of resorption lacunae made in vitro.

The assessment of in vitro osteoclastic activity has, until recently, been dependent on the analysis of organ culture experiments. We have developed a single cell resorption assay so that the resorptive function of individual osteoclasts could be studied. This paper examines the biological variation in the sizes of resorption lacunae produced by bone cell cultures derived from neonate rats and rabbits, and prehatch or hatchling chicks. Cultures were run for 24h for all species; and in addition for 48h for rat, 9 or 12 hours for rabbit and 3-7 hours for chick. The numbers of the nuclei of osteoclasts seeded on to plastic were counted for all three species. SEM stereophotogrammetry was used to measure areas, volumes, and maximum and average depths of the lacunae using specially designed instruments and software. Rat osteoclasts were smallest, and more chick osteoclasts were very large. There was a species difference in the onset of resorption and the sizes of pits produced, the chick osteoclasts being more vigorous resorbers than the rabbit ones, and the rat least so. For a given plan area, chick lacunae were deeper. There was a high correlation between area and volume. The range of maximum depths for a given area was high, however. Thus the mean of a few measurements of depths should not be used to calculate volume from area. At 24 hours, 77% of the rat, 47% of the rabbit and 28% of the chick lacunae were less than 1,000 microns 3 in volume; and 11% of the rat, 17% of the rabbit and 22% of the chick lacunae were between 1,000 and 2,000 microns 3 in volume. The mean values at 24 hours were 981, 2796, and 4582 microns 3 for rat, rabbit and chick lacunae respectively.

Beta-adrenoceptor subtype dependence of chronotropy in mouse embryonic stem cell-derived cardiomyocytes.

Cardiomyocytes derived from embryonic stem cells (ESCM) have potential both as an experimental model for investigating cardiac physiology and as a source for tissue repair. For both reasons it is important to characterise the responses of these cells, and one of the key modulators of contraction is the beta-adrenergic system. We therefore undertook a detailed study of the response of the spontaneous beating rate of ESCM to beta-adrenoceptor (betaAR) stimulation. Embryoid bodies (EBs) were generated from murine ES line E14Tg2a by the hanging drop method, followed by plating. Spontaneously beating areas were seen starting from 9-14 days after differentiation: the experiments described here were performed on EBs between developmental day 19 and 48. Beating cell layers were seeded with charcoal to allow tracking of movement by a video-edge detection system. Experiments were performed in physiological medium containing 1 mM Ca2+ at 37 degrees C. Isoprenaline (Iso) increased beating rate with an EC50 value of 52 nM. Iso (0.3 microM) increased basal rate from 67 +/- 7 beats per minute (bpm) to 138 +/- 18 bpm, P < 0.001, n = 22. At earlier developmental time points the response to Iso was not maintained through 5 min exposure; this spontaneous desensitisation only being observed before day 36. A repeat application of Iso after a wash period of 20 min produced reproducible effects on beating rate. Subtype dependence of the betaAR response was determined by comparing an initial response with a second in the presence of selective beta1- or beta2AR antagonists. In the presence of the specific beta1AR-blocker CGP 20712A (300 nM) the increase in rate with Iso was reduced from 207 +/- 42% of basal to 128 +/- 13%, P < 0.01. With the beta2AR-blocker ICI 118,551 (50 nM) there was no significant change in Iso response. Exposure to the muscarinic agonist, carbachol (10 microM), inhibited the increase in frequency mediated by isoprenaline, but had mixed stimulatory and inhibitory effects on basal rate. This study extends the characterisation of ESCM as a preparation for studying receptor pharmacology, and indicates that the beta1AR is the predominant subtype mediating increases in contraction rate in murine ESCM.

Load-induced proteoglycan orientation in bone tissue in vivo and in vitro.

Previous studies of Alcian blue-induced birefringence in adult avian cortical bone showed that a short period of intermittent loading rapidly produces an increased level of orientation of proteoglycans within the bone tissue. In the absence of further loading, this persists for over 24 hours. We have proposed that this phenomenon could provide a means for "capturing" the effects of transient strains, and so provide a persistent, constantly updated strain-related influence on osteocyte populations related to the bones' averaged recent strain history, in effect, a "strain memory" in bone tissue. In our present study, we use the Alcian blue-induced birefringence technique to demonstrate that proteoglycan orientation also occurs after intermittent loading of both cortical and cancellous mammalian bone in vivo and in vitro. We also show that the change in birefringence is proportional to the magnitude of the applied strain, and that the reorientation occurs rapidly, reaching a maximal value after only 50 loading cycles. Examination of electron micrographs of bone tissue after staining with cupromeronic blue allows direct visualization and quantification of the change in proteoglycan orientation produced by loading. This shows that intermittent loading is associated with a realignment of the proteoglycan protein cores, bringing them some 5 degrees closer to the direction of collagen fibrils in the bone matrix.